Chase

Game changer, maybe I should start... x.com/i/trending/203…

We're thrilled to be recognized by @FastCompany as a Top 10 Most Innovative Company in the Enterprise category. 🥳 Our focus is helping enterprises turn their AI goals into outcomes using technology built with the openness, trust, security, control and governance needed to remain competitive and resilient today. Check out the full list here → ibm.co/6016EyHSi

A lot of people dont actually know but @IBM has 4+ free courses on Coursera. I sorted them in to categories below, go make good use of them and boost your skills and resume. Best ones for beginners: 1. IBM Data Science coursera.org/professional-c… 2. IBM AI Engineering coursera.org/professional-c… 3. IBM Cybersecurity Analyst coursera.org/professional-c… 4. IBM Full Stack Developer coursera.org/professional-c…

Proud to be a part of @IBM I've been here for many many years and to see us constantly adapting and improving over the years is extraordinary.

Residual connections have long relied on fixed, uniform accumulation. Inspired by the duality of time and depth, we introduce Attention Residuals, replacing standard depth-wise recurrence with learned, input-dependent attention over preceding layers.

A strong weekend and a great result for the team! 💪

Make sure to join in and have a listen!

Chief Architect, Gabe Goodhart built a local AI coding assistant with Granite 4, Ollama, VS Code, and Continue. He walks through the full workflow — including the architecture and how to run it locally: ibm.biz/BdpqfH

Interested in quantum computing? With over 13 million downloads, Qiskit is the world's most popular open-source quantum computing stack for building and researching quantum algorithms. Dive deeper into Qiskit and our quantum ecosystem here: ibm.co/6010EHDuY

@horseonedragon I agree 100%, where are you working from? Would be awesome to connect more in private messages at some point - very busy currently!

@leechase99 Thanks! Happy to chat more about the cognitive approach. The key insight was that retrieval ranking shouldn't need a model call — frequency × recency is a solved problem in cognitive science. Saves latency and cost.

New research on scaling agent memory for long-horizon tasks. One of the biggest challenges with AI agents is memory. As tasks get longer and more complex, agents lose track of what they've learned, what they've tried, and what worked. This paper, from Accenture, introduces Memex(RL), a system that gives agents indexed experience memory. Instead of relying on raw context windows, agents build a structured, searchable index of past experiences and retrieve relevant memories as needed. Long-horizon agent tasks like deep research, multi-step coding, and complex planning all require persistent memory.

Researchers at UW Allen School and Stanford just ran the largest study ever on AI creative diversity. 70+ AI models were given the same open-ended questions. They all gave the same answers. They asked over 70 different LLMs the exact same open-ended questions. "Write a poem about time." "Suggest startup ideas." "Give me life advice." Questions where there is no single right answer. Questions where 10 different humans would give you 10 completely different responses. Instead, 70+ models from every major AI company converged on almost identical outputs. Different architectures. Different training data. Different companies. Same ideas. Same structures. Same metaphors. They named this phenomenon the "Artificial Hivemind." And the paper won the NeurIPS 2025 Best Paper Award, which is the highest recognition in AI research, handed to a small number of papers out of thousands of submissions.

@horseonedragon awesome man, interesting read!

Memex(RL) tackles a real problem — agents losing track over long-horizon tasks. The indexed experience approach is solid for structured retrieval. A different angle: instead of indexing experiences for explicit lookup, what if the retrieval system itself learned what matters? ACT-R activation (Anderson 1993) ranks memories by frequency × recency — a power law that naturally surfaces high-value experiences without manual indexing. Been running this in production for 30+ days: 230K retrievals, 12,510 associations emerged automatically from usage patterns. The system figured out which experiences matter without being told. github.com/tonitangpotato…

What a few days of racing it's been already! The year Wolff has built one of the best cars ever seen in F1, Ferrari seem to be rolling back the years to the Schumacher era... Mercedes Vs Ferrari is going to be one of the best battles ever!

This week's term: digital sovereignty - /ˈdɪdʒɪtəl ˈsɒvrənti/ Definition → the ability to control how your data, systems, and AI operate within a defined regulatory framework. Why it matters → digital sovereignty provides you with stronger trust in data and systems, compliance across regulatory frameworks, freedom to switch vendors, and control over cloud, infrastructure and AI.

Decentralization has happened in every domain of society - entertainment, finance, education, news, media. That’s why DeSci is much more than science, it’s a macro movement due to technology enabling this macro shift.

It's a thing of beauty!

DeSci: discipline, empowerment, science, community, integrity.

They say that if you chant “DeSci” 3 times, your upcoming week will be amazing. Just saying…